Back to Home
Retrieval-Augmented Generation (RAG)

RAG Pipeline

A RAG pipeline connects all the steps together: question, search, retrieval, and grounded answer.

This is a simplified learning demo showing how RAG works end to end.

Interactive Playground

Knowledge Base

Live Pipeline Visualization
❓ Ready. Click "Run RAG Pipeline" to see it work Step 0 of 8
❓
Question
→
📚
Chunk
Docs
→
🧮
Embeddings
→
🔍
Vector
Search
→
🎯
Retrieve
Top
→
🔀
Re-rank
→
📬
Send
Context
→
🤖
Grounded
Answer

Documents / Chunks

Context Given to AI

Run the pipeline to generate a grounded answer.

Statistics

6
Documents
6
Chunks
0
Retrieved Chunks
0
Re-ranked Chunks
0 chunks
Context Sent to AI
Not Started
Grounding Status

How It Works

❓
Question
→
📄
Documents
→
📚
Chunks
→
🧮
Embeddings
↓
🔍
Search
→
🎯
Retrieval
→
🔀
Re-ranking
→
🤖
Answer
Every earlier RAG lesson is one stage in this pipeline: chunking, embeddings, search, retrieval, re-ranking all happen in order.
The final answer only ever uses the retrieved and re-ranked chunks, nothing outside the context window.
Edit the question or documents, then run the pipeline again to see every step react live.
💡
Key Takeaway

RAG improves AI answers by giving the model relevant information before it responds.

Previous: Re-ranking Module Complete: Back to Home